China’s Secondary Privatization: Perspectives from the

Split-Share Structure Reform∗

Li Liao† Bibo Liu‡ Hao Wang§

Journal of Financial Economics, forthcoming

Abstract

The Split-Share Structure Reform granted legitimate trading rights to the state-

owned shares of listed state-owned enterprises (SOEs), opening up the gate to China’s

secondary privatization. The expectation of privatization quickly boosted SOE output,

profits, and employment, but did not change their operating efficiency and corporate

governance. The improvements to SOE performance are positively correlated to govern-

ment agents’ privatization-led incentive of increasing state-owned share value. In terms

of privatization methodology, the reform adopted a market mechanism that played an

effective information discovery role in converging the interests of the government and

private investors.

JEL Classification: G15, G18, G30.

Keywords: The Split-Share Structure Reform, privatization, market mechanism,

state-owned enterprise, government agents.

∗We thank Charles Cao, Haitao Li, Qian Sun, Honglin Wang, Chu Zhang, Hao Zhou, Dongming Zhu,Ning Zhu, and seminar participants at Tsinghua University, Xiamen University, CICF 2012 and the HKIMRConference 2013 for helpful discussions. The authors acknowledge funding support from the National NaturalScience Foundation of China (Grants No. 71232003 and 71272023).†Tsinghua University, PBC School of Finance, 43 Chengfu Road, Haidian District, Beijing 100083, China,

e-mail: liaol@pbcsf.tsinghua.edu.cn, tel: +86 10 62798215.‡Tsinghua University and CITIC Securities, Equity Capital Markets Division, CITIC Securities Tower,

48 Liangmaqiao Road, Beijing 100125, China, e-mail: lbb@citics.com, tel: +86 10 60836222.§Tsinghua University, School of Economics and Management, 318 Weilun Building, Tsinghua University,

Beijing 100084, China, e-mail: wanghao@sem.tsinghua.edu.cn, tel: +86 10 62797482.

China’s Secondary Privatization: Perspectives from the

Split-Share Structure Reform

Abstract

The Split-Share Structure Reform granted legitimate trading rights to the state-owned

shares of listed state-owned enterprises (SOEs), opening up the gate to China’s secondary

privatization. The expectation of privatization quickly boosted SOE output, profits, and

employment, but did not change their operating efficiency and corporate governance. The im-

provements to SOE performance are positively correlated to government agents’ privatization-

led incentive of increasing state-owned share value. In terms of privatization methodology,

the reform adopted a market mechanism that played an effective information discovery role

in converging the interests of the government and private investors.

JEL Classification: G15, G18, G30.

Keywords: The Split-Share Structure Reform, privatization, market mechanism, state-

owned enterprise, government agents.

1 Introduction

The split-share structure was a legacy of China’s initial share issue privatization (SIP), in

which state-owned enterprises (SOEs) went public to issue minority tradable shares to private

agents while the Chinese government remained in control by holding majority non-tradable

shares.1 Although it played a positive role in assisting China’s SOE ownership reform at

an early stage, the split-share structure jeopardized China’s continued privatization efforts

by restricting trading of state-owned shares in the secondary market. It also caused serious

corporate governance problems, encouraged speculation in the stock market, and blocked

mergers and acquisitions activities.2

In 2005, the Split-Share Structure Reform was initiated to dismantle the dual share struc-

ture by converting non-tradable shares into tradable shares.3 The reform removed any legal

and technical obstacles of transferring state-owned shares to private agents, opening up the

gate to China’s secondary privatization, which, in contrast to the initial SIP, would further

liberalize state-owned shares in full circulation. Although it had long been predicted that

the Split-Share Structure Reform would substantially change China’s corporate landscape

(Inoue, 2005), the country’s continued privatization efforts have not been comprehensively

studied, while China’s initial SIP during the 1990s received extensive research attention (Bai

et al., 1997; Lin et al., 1998; Lin, 2000; Sun and Tong, 2003). In this paper, we fill this re-

search gap by reviewing and evaluating the Split-Share Structure Reform. We identify the

1For almost all SOEs that went public before 2005, state-owned shares—together with shares held bynon-state legal persons, natural persons, and foreigners before initial public offerings (IPOs)—were restrictedfrom trading in the secondary market. Only new shares issued in IPOs and seasoned cash offerings and thosederived from tradable shares in rights offerings and stock splits were listed and tradable. By the end of 2004,the total number of RMB-denominated domestic shares (A-shares) outstanding was 714.9 billion, amongwhich 454.3 billion shares were non-tradables, 74% of which were state-owned.

2Section 2.2 presents in detail the corporate governance problems caused by the split-share structureand their reasons and consequences. The problems involve large shareholder rent seeking activities, such asrelated-party transactions, listed firms lending or guaranteeing loans to controlling shareholders. See alsoAllen et al. (2005), Huwang et al. (2006), Lin (2008), Deng et al. (2008), and Liao et al. (2011) for additionaldiscussions.

3The Split-Share Structure Reform remains an ongoing process as of November 2013. However, mostlisted firms finished the reform during 2005-2007, that is, 1,260 firms finished the reform by the end of 2007.As of November 2013, only three out the 1,315 listed firms with the split-share structure have not reformed.

1

reform’s privatization effect measured by the differential changes in post-reform operational

and stock performance of SOEs and comparable non-SOEs.4 We further trace the sources of

the privatization effect in an effort to generate policy implications for future privatization.

The evidence shows that the output, profit, and employment of listed firms increased

substantially after the reform and much more so for SOEs. Such differences were reflected in

higher SOE stock returns, consistent with Megginson et al. (1994) in that firms with better

incentives to increase stock value boost output and profits. After the reform, SOEs and

non-SOEs experienced similar degrees of increase in profitability and operating efficiency,

measured by accounts receivable turnover and expense-to-revenue ratio. Evidence on change

in corporate governance, measured by agency activities of controlling shareholders, for SOEs

and non-SOEs is however mixed. It appears that the Split-Share Structure Reform created

privatization expectation that quickly boosted SOE output and profits, but did not change

SOE corporate governance or operating efficiency without fundamentally altering their own-

ership structure.

The privatization effect is sourced to several factors. After the reform, state-owned shares

became market priced and could be conveniently transferred to private agents in the sec-

ondary market. Therefore, the interest of government agents operating and controlling SOEs

became better aligned with that of private investors.5 In particular, government agents will

be rewarded with control enhancement and favorable promotion opportunities if they improve

SOE performance, which leads to higher state-owned share value in preparation for in-depth

privatization. The evidence shows that privatization-led improvements to post-reform SOE

performance are positively correlated to government agents’ supportive activities to SOEs,

4A firm is defined in the paper as SOE if its ultimate controlling party is the state, non-SOE otherwise.See Section 4.1.1 for details. The Split-Share Structure Reform was simultaneously carried out on SOEs andcomparable non-SOEs that had the split-share structure. For non-SOEs, the reform dismantled the dualshare structure. For SOEs, the reform not only dismantled the dual share structure, but also liberalizedstate-owned shares, creating a privatization effect.

5We use “government agents” to denote executives of SOEs and their controlling shareholders. Almostall Chinese listed SOEs have state-owned controlling shareholders, who hold majority non-tradable shares onbehalf of the Chinese government. Executives of listed SOEs are appointed and evaluated by their controllingshareholders, whose executives are appointed and evaluated by the Chinese government.

2

but negatively correlated to post-reform state-owned share sales, which are considered a

punitive mechanism to government agents.

Compared to privatization in the other countries (Megginson and Neffer, 2001; Stiglitz,

2002) and failed early privatization attempts in China, the Split-Share Structure Reform

adopted a market mechanism through which government agents and private investors nego-

tiate the terms of SOE reform plans including consideration paid to the latter.6 The market

mechanism allowed government agents to communicate with private investors their incen-

tive of improving SOE performance after the reform in exchange for the latter’s agreement

to reform. We find that privatization-led improvements to post-reform SOE performance

are negatively correlated to consideration demanded, and positively correlated to private in-

vestors’ support to reform plans. The evidence suggests that the market mechanism played

an effective information discovery role in facilitating privatization in the reform.

Our finding adds new support to the notion that privatization improves the performance

of SOEs (Megginson et al., 1994; Dewenter and Malatesta, 2001; Boubakri et al., 2005; Song

et al., 2011). Moreover, expectation on privatization can stimulate managerial incentives

and boost firm performance even before actual ownership transition takes place. The find-

ing generates implications for privatization policy, which is considered a complex task for

global economies. Outcome of privatization is influenced by not only economic, political,

legal, institutional and firm-specific factors (Megginson et al., 2004) but also privatization

methodology (Perotti, 1995; Biais and Perotti, 2002).7 In particular, we demonstrate that

6Sections 2.3 and 2.4 present in detail China’s failed privatization attempts after the SIP and the marketmechanism in the Split-Share Structure Reform, respectively.

7Megginson and Neffer (2001) and Sheshinski and Lopez-Calva (2003) provide excellent reviews of theempirical and theoretical privatization literature. In particular, Brada (1996) classifies privatization methodsinto four categories: privatization through restitution, privatization through the sale of state property (directsales and SIP), mass or voucher privatization, and privatization from below. Different economies haveadopted different privatization methods and have experienced very different results. For example, Boubakriand Cosset (1998) evaluate the financial and operating performance of newly privatized firms in developingcountries, and find significant post-privatization improvements. Harper (2002) and Black et al. (2000) finddisappointing results from the Czech Republic and Russia: Firm income, profitability, and employmentsignificantly decreased after voucher privatization. Martin and Parker (1995) show that most UK firmsdid not improve their performance after privatization through asset sales after adjusting for the businesscycle effect. Boubakri and Cosset (2002) find that 79 SOEs in 21 African countries improved their output,

3

market-involved privatization approach is more effective than crude top-down privatization

orders when China enters into an in-depth reform era. The positive elements of the Split-

Share Structure Reform provide policy implications for China’s continued economic reforms.

To the best of our knowledge, this study constitutes the first effort to review the Split-

Share Structure Reform with privatization perspectives, examining its long-term effects on

SOE performance and corporate governance.8 Importantly, the reform provides a desirable

experiment setting enabling us to overcome such methodology difficulties as sample bias, data

unreliability, and changing environments that plague the empirical privatization literature

(Megginson and Neffer, 2001; Sheshinski and Lopez-Calva, 2003). As a policy event, the

reform involved almost all listed firms, both SOEs and non-SOEs, in the world’s largest

transitional economy, providing a cross-sectional sample of unprecedented scale. Publicly

disclosed high quality financial, stock, and ownership information is available before and

after the reform. These unique features allow us to measure the effects of privatization in a

clean and reliable way.

The rest of the paper is organized as follows. Section 2 reviews the Split-Share Struc-

ture Reform and its background. Section 3 develops our hypotheses. Section 4 describes

the empirical strategy. Section 5 presents and analyzes the empirical findings. Section 6

concludes.

operating efficiency, and profitability after SIP. The above results imply that privatization methods are ofcritical importance and worthy of careful examination.

8There is a growing literature studying the Split-Share Structure Reform as a special event to examinevarious corporate finance and capital market issues. Among them, the study of Li et al. (2011) findsthat consideration is significantly influenced by efficiency gain from better risk sharing. Liao et al. (2011)examine information discovery and information-based trading during post-reform lockups. Among studieson short-term market reactions and the interaction between consideration and ownership, that of Bortolottiand Beltratti (2006) reports a statistically significant 8% positive abnormal return over the reform eventwindow after adjusting for consideration requested by tradable shareholders. Lu et al. (2008) find that thepositive abnormal returns after the reform announcement are not related to consideration paid to tradableshareholders. Firth et al. (2010) report the opposite effects of state ownership and mutual fund ownershipon consideration. Huang and Zhu (2011) find that qualified foreign institutional investors help increaseconsideration. Liu et al. (2011) report a significant reduction in cash dividends after the reform, which issignificantly related to the reduction in the ownership of the largest shareholders.

4

2 Review of the Split-Share Structure Reform

This section presents the origin of the split-share structure and the problems it had caused. It

then reviews China’s failed in-depth privatization attempts before the Split-Share Structure

Reform, followed by describing the reform process.

2.1 Formation of the split-share structure

A split-share structure involves two classes of domestic A-shares with otherwise identical

rights, tradable and non-tradable, coexisting in a listed firm.9 Before the Split-Share Struc-

ture Reform, the non-tradable shares were not listed and only allowed to be transacted

through negotiations between designated parties. In contrast, the tradable shares were is-

sued to private investors and could be transacted in the secondary market.

The origin of this dual share ownership structure can be traced back to the enterprise

ownership structure reform in 1978. By then there were only two types of enterprise owner-

ship in China: SOEs, which contributed 78% of China’s industrial output, and collectives,

which were small enterprises operated by rural municipalities or urban communities. In

the early 1980s, the Chinese government carried out a series of reforms to improve the low

productivity and shrinking efficiency of the financially plagued SOEs. Those early economic

reform attempts all eventually failed, because their limited purposes of improving manage-

rial incentives and decentralizing decision making were unable to fundamentally resolve the

complex ownership structure problems inherited from the country’s planned economy.10

The Chinese government started corporatizing a selection of small and medium SOEs in

the mid-1980s and experimented by privatizing them as a core element of the second-stage

9A Chinese firm can issue several types of common shares: A-shares are common shares priced in RMBand traded on the Shanghai or Shenzhen Stock Exchange. B-shares are listed on the domestic exchanges butpriced in US dollars and H-shares are listed on the Hong Kong Stock Exchange and priced in Hong Kongdollars. A firm can also be cross-listed overseas. For example, N- and L-shares represent shares listed on theNew York Stock Exchange and London Stock Exchange, respectively.

10See Sun and Tong (2003) for a detailed review of the goals and undesirable outcomes of these reformpolicies before SIP.

5

economic reform in 1988. The founding of the Shanghai Stock Exchange and Shenzhen Stock

Exchange in 1990 marked the start of China’s SIP. SOEs went public to issue exchange-

listed tradable shares to private agents. The SIP could be best labeled as partial because it

transferred only a small portion of SOE ownership to private agents and did little to lessen

the state’s role in corporate decision making. The split-share structure was then formed.

Transaction of state-owned shares was not an issue by then. State-owned shares, to-

gether with shares issued to legal persons, natural persons, and foreigners before IPOs, were

restricted from trading in the secondary market. This restriction was explicitly written in

IPO prospectuses or publicly announced. Only new shares issued in IPOs and seasoned cash

offerings and those derived from tradable shares in rights offerings and stock splits were trad-

able on the stock exchanges. The Tentative Measures for the Administration of the Issuance

and Trading of Stocks, issued by the State Council in April 1993, required transactions of

state-owned shares to be approved by the relevant authorities but provided no applicable

measures on implementation.

The Chinese government chose to put the state-owned share transaction issue on hold

indefinitely for several reasons. First, transactions of state-owned shares appeared unnec-

essary in a centralized ownership framework and the designated administrative system.11

Second, in the 1990s, the economic reform was still focused on the administration and man-

agement of the SOEs that went public mainly to raise capital and to experiment with the

new government-controlled management mechanism. Third, the Chinese stock market was

at an experimental stage and not ready to facilitate transactions of state-owned shares.

11By then socialism was ideologically interpreted to mean that all assets belong to all citizens. Thestate possesses these assets on behalf of the citizens at the primitive stage of socialism. The central gov-ernment represents the state in exercising ultimate control and administration of the state-owned assets.The state-owned shareholders consisted of different levels and departments of the Chinese government andtheir affiliates, who delegated the central government to manage SOEs according to their administrativefunctions. Transfers of state-owned assets due to reorganization and industry restructure were almost allexecuted through administrative orders without monetary transactions.

6

2.2 Problems caused by the split-share structure

The legacy dual share ownership structure created critical problems in the functioning and

development of China’s financial markets in the most recent years and caused tremendous

concerns. Under the split-share structure, the interests of tradable and dominant non-

tradable shareholders were fundamentally divided due to different share pricing mechanisms.

The non-tradable shares were priced according to the book value of firm assets instead of

firm market values. Controlling non-tradable shareholders, who on average possessed two-

thirds of total shares outstanding, were unable to benefit from capital gains. Hence, they

had little, if not zero, incentive to improve firm performance.

In the absence of effective internal and external monitoring,12 controlling shareholders

made listed firms raise money relentlessly through seasoned cash offerings, ignoring adverse

market reactions and control dilution because of their absolute dominance. After raising

money, controlling shareholders duly extracted value through related-party transactions,

where controlling non-tradable shareholders transferred wealth through transactions, such

as asset sales and product purchases, with entities they owned (before the reform 29.7%

of firms in our sample engaged in related-party transactions with controlling shareholders),

corporate lending, where listed firms made loans to controlling shareholders (before the

reform 42.3% of firms in our sample made loans to their controlling shareholders), and listed

firms guaranteeing loans to controlling shareholders.

In equilibrium, investors speculated in the stock market for short-term returns rather

than invested for long-term capital gains. Overtrading was rampant in the Chinese stock

market. As of 2007, the average turnover ratios of the Shanghai and Shenzhen Stock Ex-

changes were 927% and 987%, respectively. In comparison, as of 2005, the average turnover

ratios of the stock markets in the US, UK, and Japan were 129%, 142%, and 119%, respec-

12Boards of directors nominated by dominant non-tradable shareholders did not function effectively interms of monitoring managerial behaviors in the best interests of minority tradable shareholders. Externalmonitoring through corporate takeovers was not feasible due to non-transferability of dominant non-tradableshares.

7

tively (China Capital Markets Development Report, China Securities Regulatory Commission

(CSRC), 2008). Such high turnover ratios cannot be justified by liquidity, which should be

accompanied by low volatility of stock returns. Liao et al. (2010) show that between 1995

and 2008 the average monthly stock return volatilities of the Shenzhen and Shanghai Stock

Exchanges were 10.7% and 8.9%, respectively, the highest in the world, and 19 and 16 times

the average stock return volatility of the New York Stock Exchange. Highly volatile stock

prices are likely driven by Chinese investors’ speculation activities (Sun and Tong, 2003;

Allen et al., 2005; Xiong and Yu, 2011).

The lack of controlling shareholder incentive to finance with debt to avoid financial dis-

tress together with the corporate bond pricing difficulties introduced by the split-share struc-

ture fundamentally discouraged the development of domestic corporate debt and derivative

markets. As of 2007, China’s bond market capitalization to gross domestic product (GDP)

ratio was 35.3%, far lower than the 188.5% and 201.0% of the US and Japan, respectively.

Corporate bonds amounted to only 4.2% of China’s bond market. The ratio of outstanding

corporate bonds to GDP was 1.5% for China, in comparison to 125.7% and 38.9% for the

US and Japan, respectively (China Capital Markets Development Report, 2008).

2.3 Failed early privatization attempts

The Split-Share Structure Reform, as part of China’s in-depth privatization efforts, did not

come easily. This section presents China’s failed privatization attempts after the initial SIP

and their reasons.

In September 1999, the Fourth Plenum of the 15th Central Committee of the Communist

Party of China passed the Decision of the Central Committee of the Communist Party of

China on Major Issues Concerning the Reform and Development of State-Owned Enterprises.

The decision aimed to privatize an unspecified number of state-owned shares to raise capital

for the Social Security Fund, which served to lessen the welfare burden of SOEs. The

privatization, however, was conditional on the state securing absolute control of those SOEs

8

to be further privatized.

In December 1999, the CSRC handpicked ten listed companies to pilot the state-owned

share sales. As in rights offerings, preferential subscription rights were offered to existing

shareholders. Sale prices were crudely set using firms’ average earnings per share in prior

three years multiplied by a fixed price earnings ratio of ten. Sales were quickly suspended

after trying two companies because the stock market reacted negatively due to the dramatic

discrepancy between the sales prices and prices the market was expecting. The Shanghai

Composite Index and the Shenzhen Composite Index dropped by 7.3% and 6.8%, respec-

tively, during the 25-day sale period.

On June 12, 2001 the State Council issued the Interim Measures of the State Council on

the Management of Reducing State Shares Held and Raising Social Security Funds, stating

that SOEs would privatize 10% of state-owned shares in IPOs and seasoned cash offerings.

The price of state-owned shares would be set equal to the offering price of newly issued

shares. Sales were halted on October 22, 2001 after 16 SOEs practiced the interim mea-

sures, thus inviting tremendous adverse market reaction. During four months, the Shanghai

and Shenzhen Composite Indexes plummeted by 31.0% and 32.9%, respectively. The stock

market remained bearish throughout 2002-2004 with transaction volumes shrunk by nearly

30%. The Shanghai Composite Index plunged from a record high of 2,245 points on June 14,

2001 to 998 points on June 6, 2005. During the same period of time, the Chinese economy

experienced 11% annual compounding growth.

Why did the market react so adversely? Besides the market being concerned that the

rapidly inflated stock volume might flood the secondary market, a more fundamental reason

was that these privatization attempts directly breached IPO and seasoned equity offering

agreements on non-tradability of state-owned shares. Privatization harmed investor interests

but provided no compensation. That created widespread dissatisfaction as well as anxiety

over the overhung state-owned shares. In November 2001 the CSRC solicited public opinions

and suggestions on practical method to privatize state-owned shares. No satisfactory reso-

9

lution was reached because the market refused to accept the idea of privatizing state-owned

shares without systematically legitimizing their trading rights and compensating tradable

shareholders.13

2.4 The Split-Share Structure Reform

The Chinese government gradually realized that in-depth privatization and market liber-

alization could not be accomplished without completely dismantling the legacy dual share

structure inherited from the SIP. The Split-Share Reform was then initiated to liberalize

state-owned shares in full circulation. On January 31, 2004 the State Council issued Some

Opinions of the State Council on Promoting the Reform, Opening and Steady Growth of

Capital Markets as a blueprint of the reform. The Notice of the China Securities Regulatory

Commission on Piloting the Share-Trading Reform of Listed Companies issued on April 30,

2005 marked the official start of the Split-Share Structure Reform.14 Instead of directly sell-

ing state-owned shares, the reform aimed to convert all non-tradable shares into legitimate

tradable shares with negotiated considerations to compensate tradable shareholders. To en-

courage listed firms to reform, the CSRC imposed the reform as a prerequisite for seasoned

equity offerings.

A firm’s reform process typically has several steps. First, if at least two-thirds of non-

tradable shareholders agree to reform, the board of directors authorizes the management to

hire a qualified securities firm as facilitator to seek an agreement with the domestic exchange

on which the firm is listed on a tentative reform schedule. Non-tradable shareholders then

13Other privatization methods, including contract-based transaction of state-owned shares, state-ownedshare-to-debt swaps, and auctions, were either considered or pilot-tested but quickly withdrawn. For example,in January 2003 the CSRC announced a plan to sell state-owned shares, together with other non-tradableshares, at discounted prices. Sale price would be determined through public auction and below the secondarymarket price. After sales, non-tradable shareholders would compensate tradable shareholders through sharetransfers or designated rights offerings. The plan was withdrawn in two days after the Shanghai CompositeIndex lost 6%.

14Four companies, namely, Sany Heavy Industry, Tongfang Co., Zijiang Enterprise Group, and JinniuEnergy Resources, comprised the first batch of pilot firms chosen by the CSRC. The second pilot batchincluded 42 companies.

10

propose a reform plan with consideration paid to tradable shareholders in exchange for the

trading rights of non-tradable shares. The consideration can be paid in cash, stock, stock

option, or warrant.15 Controlling non-tradable shareholders can make written promises on

future dividend payouts and/or asset injections to sweeten the consideration package. The

proposal is then circulated for tradable shareholder feedback and negotiation.16

The reform adopts a market mechanism through which tradable and non-tradable share-

holders negotiate the terms of reform plan. On the principle of ‘fair negotiation, mutual trust,

and independent decision making’ (China Capital Markets Development Report, 2008), the

negotiation reflects each firm’s specific situation. No government intervention or standard

pricing is imposed. After receiving positive feedback from tradable shareholders during ne-

gotiation, the firm calls for a special shareholder meeting in which tradable shareholders

vote to approve the proposal. The firm is required to provide the necessary information

technology system for tradable shareholders to vote online. Trading of tradable shares is

frozen on the meeting day. The reform plan is submitted to the CSRC for final approval if

more than two-thirds of participating tradable shareholders approve it. After the CSRC’s

approval, the reform plan becomes effective. Trading of tradable shares resumes the next

trading day.

To stabilize the stock market, each firm’s reform plan contains a compulsory non-tradable

share lockup of 12 months after the reform plan’s effective day. In addition, a non-tradable

shareholder is restricted from selling more than 5% (10%) of total shares outstanding within

12 (24) months after the lockup. Transactions of non-tradable shares over 1% of total shares

outstanding must be publicly disclosed within two trading days. Upon completion of the

15Li et al. (2011) report that the average (median) value of consideration, measured as the number ofshares transferred to tradable shareholders for each tradable share held is 0.305 (0.310) for firms whosenon-tradable shareholders paid considerations in stock only.

16It was not uncommon for a negotiation to take several rounds. For example, Tongfang Co., one of thefirst batch of pilot firms, disregarded the negative feedback from tradable shareholders and held a specialshareholder meeting in which its reform plan proposal was rejected and returned for re-proposal. Thatsubstantially delayed the company’s reform. Later on, all companies tended to renegotiate and re-proposetheir plans after learning of tradable shareholder dissatisfaction.

11

Split-Share Structure Reform, the Chinese stock market would no longer be fundamentally

different from international markets in terms of pricing and valuation.

3 Hypothesis development

This section develops hypotheses to identify the privatization effect embedded in the Split-

Share Structure reform, and determines the sources of the privatization effect. Theory pre-

dicts that private ownership is more efficient than government ownership because a competi-

tive equilibrium is Pareto optimal (Megginson and Neffer, 2001). Sheshinski and Lopez-Calva

(2003) argue that significant efficiency gain should be obtained for firms being transferred

from government to private ownership in competitive industries. Empirical evidence indi-

cates that in many economies and industries, firm performance is improved after privatization

(Megginson et al., 1994; LaPorta et al., 1999; Boubakri et al., 2005). We develop and test

the following hypothesis on the existence of a privatization effect in the Split-Share Structure

Reform:

H1: After the Split-Share Structure Reform, SOEs improved performance more than non-

SOEs.

The next question is on the sources of such privatization effect, if exists. Megginson

et al. (1994) show that stronger managerial incentive of increasing share value leads to

increases in SOE output and profit. The interest of the Chinese government and the incentive

of the government agents running SOEs play central roles in shaping SOE performance

and corporate governance, due to the different corporate structure of Chinese listed SOEs

from that of western firms. Almost all listed SOEs in China have state-owned controlling

shareholders, who hold majority non-tradable shares on behalf of the Chinese government.

Executives of listed SOEs are appointed and evaluated by their controlling shareholders,

whose executives are appointed and evaluated by the Chinese government.

Before the reform, the split-share structure of listed firms caused serious governance

problems and is regarded as a failure of China’s SIP (Allen et al., 2005; Deng et al., 2008).

12

Jensen and Meckling (1976) state that a particular form of agency problem involves the

interest conflicts between controlling shareholders and minority shareholders in a market

with high ownership concentration. Grossman and Hart (1988) and Shleifer and Vishny

(1997) show that controlling shareholders have incentives to divert firm resources for private

interest at the expense of minority shareholders. Since private investors hold tradable shares

of SOEs, the agency problems of SOE controlling shareholders are rooted in the interest

conflicts between government agents and private investors.

The Split-Share Structure Reform generated expectation of further privatization.17 After

the reform, state-owned shares became market priced and could be easily transferred to pri-

vate agents in the secondary market. Performance of government agents will be evaluated on

the market values of SOE shares, instead of the book values of SOE assets as before the re-

form. Hence, the interests of government agents and private investors become better aligned.

In particular, government agents will be rewarded with control enhancement and favorable

promotion opportunities if they improve SOE performance and increase state-owned share

values in preparation for privatization. For example, good SOE performance could help their

controlling shareholders become group listed.18 On the other hand, the government could

divest in underperforming SOEs to discipline their management. Sales of state-owned shares

constitute a punitive mechanism that dilutes government agents’ control and jeopardizes

their future promotion. Furthermore, SOEs with good performance could have better access

to seasoned equity offerings, which are subject to profitability requirements. They could

17As shown in the previous section, the Chinese government had attempted but failed to privatize state-owned shares of listed SOEs to raise capital for in-depth economic reform. Although it remains unclearwhen and how the Chinese government will privatize SOEs after the Split-Share Structure Reform, whichhas nevertheless removed the legal and technical obstacles. The reform also pointed out the direction forfuture policies. Rapidly growing government liabilities and loss of SOE competitiveness after further financialliberalization are potential triggers for large-scale privatization. Consistently, the third plenum of the 18thCentral Committee of the Chinese Communist Party closed on November 12, 2013 decides to allow moreprivate capital into the market to develop a mixed ownership economy.

18On August 23, 2005 the CSRC, jointed by the State-owned Assets Supervision and AdministrationCommission of the State Council, the Ministry of Finance, the People’s Bank of China (the central bank),and the Ministry of Commerce, announced The Guiding Opinions on the Split-Share Structure Reform ofListed Companies stating that the government encourages enterprises with good performance to becomegroup-listed through the SOEs under their control after the Split-Share Structure Reform.

13

also enjoy lower financing costs, because higher firm value derives lower insolvency risk and

better quality collateral. Given that privatization-led incentive of government agents reflects

in their supportive activities to SOEs, we develop and test the following hypothesis:

H2: Privatization-driven improvements to post-reform SOE performance are positively

correlated to government agents’ supportive activities.

We further examine the information discovery role of the market mechanism in facilitating

privatization, arguing that the market mechanism helps to bridge information exchange be-

tween government agents and private investors at the reform negotiation stage. Government

agents offer either up-front consideration or future capital gain in exchange for the latter’s

agreement to reform. If the market mechanism is effective in facilitating the information ex-

change, private investors should be able to form reasonable expectation on the post-reform

performance of SOEs, which in turn affects their approval of reform plans and the amount

of considerations demanded. Therefore, we develop and test the following hypotheses:

H3: Privatization-driven improvements to post-reform SOE performance are negatively

correlated to consideration demanded and positively correlated to private investors’ reform

plan approval rate.

4 Empirical strategy

This section presents our empirical strategy, first illustrating the methods used to identify

the privatization effect, followed by introducing the data. It then introduces how to relate

the privatization effect to several potential sources.

4.1 Identifying the privatization effect

Our greatest advantage in studying privatization is that the Split-Share Structure Reform

was simultaneously carried out on SOEs and comparable non-SOEs with the split-share

structure. For non-SOEs, the reform dismantled the dual share structure. For SOEs, the

reform not only dismantled the dual share structure, but also removed legal and technical ob-

14

stacles restricting transfer of state ownership to private agents. That generated expectation

for in-depth privatization, creating a privatization effect on SOEs.

4.1.1 Measuring state ownership

For robustness we measure state ownership in three ways: (1) We define a firm as SOE if its

ultimate controlling party is the state, non-SOE otherwise. Chinese listed firms disclose their

ultimate controlling parties in annual financial reports. The state is the ultimate controlling

party of a firm if the state controls directly or indirectly over 50% of total shares outstanding,

the state controls directly or indirectly over 30% of total voting rights, the state’s voting

rights can elect over 50% of board directors, or the state has significant influence on decisions

made in shareholder meetings; (2) we use the ratio of state-owned shares to total shares

outstanding as a proxy for state-ownership in regressions; and (3) we divide firms by state

ownership measured in (2) into four groups, where Group P contains private listed firms

that do not have any state-owned shares, constituting a sub-sample of the non-SOEs.19 The

rest of the firms are ranked by their ratios of state-owned shares to total shares outstanding

from low to high, and assigned to Groups SL, SM , and SH , respectively.

4.1.2 Evaluating the reform and estimating the privatization effect

We contrast firm operating performance and corporate governance three years before and

after the reform in evaluating the success of the reform. Medians, instead of means, of the

variables are examined, because, as for Sun and Tong (2003), we note that the distributions

of the key variables of interest are heavily skewed and leptokurtic.20 As a result, we apply

Wilcoxon signed-rank tests to examine the significance of the median changes in the variables

19We manually trace the origins of the 146 private listed firms, and find that nearly 43% of the firms havenever been state-owned. Their shares held by non-state legal persons, natural persons, and foreigners beforeIPOs were restricted from trading in the secondary market before the Split-Share Structure Reform. About53% of the firms had state-owned shares at IPOs. The state-owned shares were transferred to private agentson individual case base as China continued experimenting privatization methodology after the SIP. Theseprivatized shares remained untradable before the Split-Share Structure Reform.

20For example, the Shapiro-Wilk test on operating revenue generates W=0.13 with p<0.0001, rejectingthe null hypothesis that the variable is normally distributed.

15

before and after the reform, and Wilcoxon Z-tests to examine the significance of the difference

in the median changes between groups.

Following Megginson et al. (1994), Dewenter and Malatesta (2001), and Sun and Tong

(2003), we adopt a difference-in-difference approach to measure the privatization effect. We

first divide the non-SOEs into 5x5 benchmark portfolios by size (measured by market capital-

ization) and industry. For robustness, we construct another set of 5x5 non-SOE benchmark

portfolios by size and market-to-book ratio. We then assign each SOE to one of the 5x5

benchmark portfolios by matching size and industry (size and market-to-book ratio). We

compute the median changes in the performance and corporate governance variables of these

benchmark portfolios, ∆PerformBK , and those of each SOE, ∆PerformSOE, respectively.

We compute in the final step the privatization-led improvements to SOE performance as

IMP PRIV = ∆PerformSOE −∆PerformBK , removing the non-privatization effects of the

reform and influence of unknown economic shocks.

4.1.3 Operational performance

It is important to consider that China revised Chinese Generally Accepted Accounting Princi-

ples (GAAP) to embrace the International Financial Reporting Standards during the sample

period. The new Chinese GAAP became effective in January 2007 with revisions centering

on how investment profit and other income are scoped and calculated. Assets and share-

holder equity are affected by the changes in calculating inventories and retained earnings

(Ding and Su, 2008; Peng and Bewley, 2010). The revision renders incomparable earnings-

and asset-based financial measures, such as return on assets (ROA) and return on equity

(ROE), before and after the reform.

As a result, we use the Consumer Price Index-adjusted operating revenue and operating

profit as proxies for output, the number of employees and capital expenditure (measured

as change in gross property, plant, and equipment (PPE) plus change in intangible assets)

normalized by operating revenue as proxies for employment and investment, respectively.

16

We use operating revenue per employee and operating profit per employee as proxies for

productivity and profitability, respectively. We compute the accounts receivable turnover

and the ratio of selling and financial expenses to operating revenue to measure operating

efficiency.21 Since the accounting variables used in computing the frequently used debt-to-

asset and current ratios are also affected by the revision of the Chinese GAAP, we use the

ratio of cash to total liabilities as a proxy of insolvency risk. These common but incomparable

financial measures, such as ROA, ROE, and debt-to-asset ratio, are also reported only for

reference.

4.1.4 Stock market performance

To measure firm market performance, we estimate the Fama and French (1993) three-factor

model-adjusted stock returns, and compute the model-implied stock returns with

r̂i = βMi (rM − rf ) + βSMB

i rSMB + βHMLi rHML + rf , (1)

where rM , rf , rSMB, and rHML denote the market (Shanghai Composite Index) return, the

risk-free (one year Chinese Treasury) rate, and the returns for the SMB and HML portfolios,

respectively. β is estimated using one-year pre-reform daily stock returns. We compute

adjusted stock return as radj = ri − r̂i, where ri denotes raw stock return. Privatization-

driven SOE stock return is computed as rPRIVadj = rSOE

adj − rBKadj , where rSOE

adj denotes an SOE’s

adjusted stock return, and rBKadj denotes the median adjusted stock return of the non-SOE

benchmark portfolio, to which the SOE is assigned by matching size and industry (or size

and market-to-book ratio).

21General and administrative expenses are affected by the revision of the Chinese GAAP due to suchchanges as in calculating inventories and assets. Selling and financial expenses are not affected by therevision, and applied as a substitute for commonly used selling, general, and administrative expenses.

17

4.1.5 Corporate governance

Traditional corporate governance tools, such as internal monitoring and external takeover,

have been shown ineffective for listed Chinese firms (Sun and Tong, 2003; Allen et al., 2005).

We propose three novel measures of corporate governance: (1) related-party transactions,

(2) related-party transactions with controlling shareholders, and (3) listed firms lending

to controlling shareholders. These activities represented notorious controlling shareholder

agency problems before the reform (Xu et al., 2005; Ma et al., 2005; Cheung et al., 2006;

Hou et al., 2008). We would like to examine whether the percentage of firms engaged and

the amount of money involved in these agency activities change before and after the reform,

in assessing the effect of the reform on listed firms’ corporate governance.

4.2 Data

Data on the Split-Share Structure Reform and firm financial information are obtained from

the CSMAR database and cross-checked against data in the WIND database to improve

reliability. Firm operating revenues before the revision of the Chinese GAAP are manually

collected from firm annual reports, since they are not reported in the databases. There are

1,260 firms completed the Split-Share Structure Reform by the end of 2007. We exclude 228

firms that are delisted, in the financial industry, or with incomplete accounting information.

Our final sample contains 1,032 firms, among which 633 are SOEs and 399 are non-SOEs.

Groups P , SL, SM , and SH contain 146, 295, 296, and 295 firms, respectively.

Table 1 reports the medians of a selection of key variables one year before the reform being

effective. The medians of the ratios of non-tradable shares to total shares outstanding for

non-SOEs and SOEs are similar, at 60.15% and 62.51%, respectively. The SOEs, originating

from capital-intensive heavy industries, are larger than the non-SOEs in terms of assets,

revenue, and employment. For example, the non-SOEs’ median assets and annual revenue

are RMB 1,264.25 million and 592.31 million, respectively. The SOEs’ were RMB 2,069.67

18

million and 1,236.01 million, respectively.22 The SOEs and non-SOEs have similar levels

of profitability. The medians of the SOE (non-SOE) net margin rate, ROA, and ROE are

3.50% (3.66%), 2.50% (2.08%), and 5.26% (4.25%), respectively. The SOEs and non-SOEs

have comparable leverage ratios and short-term solvency measures, that is, the medians of

the SOE (non-SOE) debt-to-asset ratios and current ratios are 0.51 (0.55) and 1.17(1.14),

respectively. The market-to-book ratios for the SOEs and non-SOEs are 1.70 and 1.76,

respectively, comparable to the average market-to-book ratio of US firms.23 The SOEs have

a median accounts receivable turnover of 5.97, higher than the 3.53 for the non-SOEs.

Managerial ownership is low for listed Chinese firms. The median managerial share-

holding is 0.004% for the full sample, that is, average firm management holds four out

of 100,000 shares outstanding. In general, non-SOE managers hold higher percentage of

shares outstanding than SOE managers. The median managerial shareholding for SOEs

and non-SOEs are 0.004% and 0.005%, respectively. The Chinese government restricts SOE

managerial shareholding in avoiding value and control dilution of state assets. We, there-

fore, do not use management shareholding as a primary managerial incentive measure for

government agents running SOEs.

4.3 Sources of the privatization effect

This section first examines the significance of privatization effect in the reform, and then

applies regression analysis to relate the privatization effect to potential sources.

22The RMB appreciated steadily against the US dollar during our sample period. The average exchangerates of RMB to US dollars were 8.28, 8.29, 7.97, 7.60, 6.95, and 6.83 from 2004 to 2009, respectively. Onaverage, US$1 was equivalent to RMB 7.64 during our sample period.

23Pontiff and Schall (1998) report an average book-to-market ratio of 0.668 for US firms during 1926-1994, which translates into an average market-to-book ratio of 1.50. Chen et al. (2013) report an averagemarket-to-book ratio of 2.13 for US firms during 1950-2010.

19

4.3.1 Regression setup

We test H1 using the following regression with the full sample:

∆Performi = αi + βStateOwnStateOwni +N∑j=1

βControlj Controli,j + εi, (2)

where ∆Performi denotes the change in firm i’s operational performance or stock return.

StateOwni denotes state ownership measured by the ratio of number of state-owned shares

to number of total shares outstanding. According to H1, the coefficient of StateOwn is

expected to be positive.

We include the ratio of non-tradable shares to tradable shares in controlling for the

relative bargaining power of non-tradable shareholders in reform plan negotiation, where the

higher the ratio, the stronger the relative bargaining power of non-tradable shareholders.

There could be concerns that stronger improvements to SOE operating performance are

driven by their monopoly power (Megginson and Neffer, 2001). To address the issue, we

include the logarithm of market equity value and a regulated industry dummy to control for

such an effect, because large SOEs tend to possess the strongest monopoly power in regulated

industries, such as telecommunications and natural resources (Sun and Tong, 2003). We

include a Hong Kong cross-listing dummy in the regressions to control for the cross-listing

effect. We control for the year effect by including dummies for the years 2005 and 2006,

respectively.

We investigate whether improvements to post-reform SOE performance are positively

correlated to controlling shareholders’ supportive activities with the SOE sample and the

following regression:

IMP PRIVi = αi + βGroupListGroupListi + βAssetInjectAssetInjecti

+βFundRaiseFundRaisei +N∑j=1

βControlj Controli,j + εi, (3)

20

where IMP PRIVi represents privatization-driven improvements to SOE post-reform perfor-

mance, measured by operating revenue, operating profit, and stock return, respectively.

CroupList represents the group-listing dummy taking 1 if the controlling shareholder of an

SOE became group-listed after the reform—majority assets of the controlling shareholders

were injected into the SOE, and 0 otherwise. AssetInject represents the asset injection

dummy taking 1 if the controlling shareholder of an SOE injected assets into the SOE after

the reform, and 0 otherwise. FundRaise represents the number of rounds of external fund

raising after the reform. According to H2, CroupListi, AssetInjecti, and FundRaisei should

all be positively correlated to IMP PRIVi . To avoid the multicollinearity problem, we include

these supportive activities each individually as independent variables in the regressions.

To analyze whether the market mechanism played an effective informational role in facil-

itating privatization embedded in the reform, we carry out cross-sectional regressions based

on the following equation with the SOE sample to test H3:

IMP PRIVi = αi + βConsiderConsideri + βApprovalApprovali +

N∑j=1

βControlj Controli,j + εi, (4)

where Consider and Approval denote consideration paid to tradable shareholders and trad-

able shareholder reform plan approval rate, respectively. Consider is measured as the ratio

of number of shares transferred from non-tradable shareholders to tradable shareholders to

number of tradable shares outstanding (Li et al., 2011). Tradable shareholders’ reform plan

approval rate is computed as the ratio of number of tradable shares voted to approve the re-

form plan to number of tradable shares participated in the vote. According to H3, Consideri,t

and Approvali should be negatively and positively correlated to IMP PRIVi, , respectively.

The independent variables and the residuals in the above equations are not normally

distributed.24 In such case, ordinary least square (OLS) regressions can produce biased

24For example, the skewness and kurtosis of the residuals in the operating revenue regression specified inEq. (2) are 3.3 and 15.8, respectively. The Shapiro-Wilk test on the residuals shows W=0.70 with p<0.0001,rejecting the null hypothesis that the residuals are normally distributed.

21

estimates and unreliable statistical inferences. Thus, we apply quantile regressions in our

investigation (Koenker and Bassett, 1978). Quantile regression imposes no restrictive prior

on the distribution of residuals, and utilizes the least absolute distance estimation algorithm

instead of the least-squares algorithm. It allows one to examine any arbitrary quantiles of

selected dependent variables, enabling the investigation on the significance and stability of

results all over a quantile spectrum. We analyze the most representative 25%, 50% (median)

and 75% quantile, and illustrate the robustness of results over the entire quantile range.

5 Result analysis

This section evaluates the Split-Share Structure Reform and its privatization effect look-

ing into the post-reform changes in SOE and non-SOE output, employment, productivity,

operating efficiency, and corporate governance. We find that the listed firms substantially

increased output, profit, and employment after the reform, with SOEs significantly outper-

forming their counterparts. Such differences are duly reflected in their higher stock returns.

There is no consistent evidence that expectation of privatization led to greater improvements

in operating efficiency and corporate governance for SOEs than non-SOEs. Improvements

to SOE performance are positively correlated to government agents’ incentive of increasing

state-owned share value. The market mechanism plays a remarkable information discovery

role in facilitating privatization.

5.1 Evaluating the reform and privatization effect

5.1.1 Output and productivity

Table 2 reports the post-reform changes in firm output, employment, and investment. It

shows that firm operating revenue increases significantly after the reform. For the full sample,

the median increase is 73%, significant at the 1% level. The result is unlikely driven by

outliers as Column (1) shows that 779 firms experienced positive operating revenue change,

22

whereas 253 firms experienced downward change. The increases in median operating revenue

for the non-SOEs and SOEs are 57% and 84%, respectively. Moreover, the increases in

the medians for groups P , SL, SM and SH were 47%, 60%, 74%, and 92%, respectively,

significant at the 1% level. SOEs experienced greater output growth than non-SOEs, that

is, the difference in median output growth between SOEs and non-SOEs (the highest state

ownership group SH and the pure private group P ) was 27% (44%), significant at the 1%

level. Expectation of privatization can have two opposite effects on output. On the one

hand, there is a positive effect due to better-aligned government agent incentive and more

flexible financing (Megginson et al., 1994). On the other hand, output could decrease due to

a reduction in government subsidy (Boycko et al., 1996). The first effect appears to dominate

the second in the expectation of privatization in our case.

The evidence indicates significant post-reform increases in firm operating profit. The

median increases for the full sample, non-SOEs, and SOEs are 45%, 44%, and 50%, respec-

tively, significant at the 1% level. The difference between the SOEs and non-SOEs is however

insignificant. Employment increased substantially after the reform with a median change for

the full sample of 13%, significant at the 1% level. The difference in the changes of the

median employment growth rate between SOEs and non-SOEs (Group SH and Group P ) is

17% (16%), significant at the 1% level. Higher growth in SOE employment after the reform

can not be entirely due to new positions created. Some employees could have been trans-

ferred from the SOE’s controlling shareholders with asset injections. The reform nonetheless

did not cause mass reduction in employment, which is in stark contrast to China’s SIP that

led to sizable unemployment in the 1990s (Sun and Tong, 2003).

Post-reform growth in capital asset investment was slower than operating revenue growth.

The changes in the median ratios of capital expenditure to operating revenue for the full

sample, non-SOEs, and SOEs are -2.76%, -3.87%, -2.04%, respectively. SOEs experienced

lower decreases in capital assets relative to their operating revenues than non-SOEs. The

difference in the changes of the median ratios of capital expenditure to operating revenue

23

between SOEs and non-SOEs (Group SH and Group P ) is 1.83% (5.16%), significant at the

10% (1%) level. The result is consistent with H2 in that government agents had stronger

incentive to inject high-quality assets into SOEs in boosting their performance and increasing

stock prices after the reform.

Theory and empirical evidence show that a dual share structure misaligns the control

and cashflow rights of controlling shareholders, and negatively affects productivity and prof-

itability (Jensen and Meckling, 1976; Gompers et al., 2008; Masulis et al., 2009). Panel B of

Table 3 reports the changes in operating revenue per employee and profit per employee as

proxies for productivity. Operating revenue (profit) per employee increased significantly, by

41% (17%), 41% (19%), and 41% (15%) for the full sample, non-SOEs, and SOEs, respec-

tively. However, there is no significant difference between SOEs and non-SOEs (Group SH

and Group P ).

5.1.2 Operating efficiency and insolvency risk

Table 3 reports the post-reform changes in firm operating efficiency and insolvency risk. For

operating efficiency, the increases in the median accounts receivable turnover for all firms,

non-SOEs, and SOEs are 4.51, 4.24, and 4.64 times, respectively, significant at the 1% level.

The increases in the median accounts receivable turnover for groups P , SL, SM , and SH are

3.78, 5.26, 4.46 and 4.60 times, respectively, significant at the 1% level. Firms on the one

hand experienced efficiency gains, and on the other hand became more discreet with credit

sales, particularly during the global financial crisis in 2008-2009.25 The difference between

SOEs and non-SOEs is however insignificant. The ratio of selling and financial expenses

to operating revenue exhibits similar patterns for SOEs and non-SOEs with an increase of

3%. We find mixed evidence on improvements in operating efficiency with no significant

difference between SOEs and non-SOEs. Panel B of Table 3 reports changes in leverage and

25The full sample median accounts receivable turnover three years before the reform is 4.40. The medianpost-reform increase in operating revenue is 73%, while the median decrease in accounts receivable is 13%.These results imply a median post-reform accounts receivable turnover of 8.70, leading to a net change of4.30, roughly matching the actual change of 4.51.

24

solvency proxies after the reform. The ratio of cash to total liabilities displays no significant

change in listed firms’ insolvency risk.

5.1.3 Corporate governance

Our corporate governance measures include (1) related-party transactions, (2) related-party

transactions with controlling shareholders, and (3) listed firm lending to controlling share-

holders. Related-party transactions, especially those involving controlling shareholders, were

accused of being a popular way of profit tunneling before the reform (Cheung et al., 2006).

Controlling shareholder borrowing at extremely low cost or even interest free from listed

firms has also been widely criticized (Ma et al., 2005; Hou et al., 2008).

Panel A of Table 4 reports the percentage of firms engaging in such activities. Before

the reform, there were 43.4%, 29.7%, and 42.3% of firms in our sample engaged in related-

party transactions, related-party transactions with controlling shareholders, and lending to

controlling shareholders, respectively. After the reform, the percentages dropped to 35.7%,

24.0%, and 16.6%, respectively. The percentage of firms engaged in these agency activities

increases monotonically with state ownership. After the reform, the percentage of SH firms

engaged in related-party transactions decreased more than the private listed firms. However,

the percentage of SH firms engaged in related-party transactions with controlling sharehold-

ers and lending to controlling shareholders reduced less compared to firms in the private

group. Panel B of Table 4 reports that the changes in the ratios of the amount of fund

involved in these agency activities to operating revenue are insignificant. Overall, there is no

consistent evidence that SOEs experienced greater improvements in corporate governance

than the non-SOEs, suggesting that expectation of privatization might quickly boost SOE

output and profit, but did not change corporate governance.

25

5.1.4 The privatization effect

Table 5 summarizes the privatization effect measured by changes in SOE operating revenue

and operating profit, and Fama-French model-adjusted stock returns, respectively. An av-

erage SOE’s operating revenue and operating profit increase by 84% and 50%, respectively.

Its adjusted stock return is 109%. The changes are statistically significant at the 1% level.

SOEs exhibit 6% (6%) higher increase in operating revenue, 2% (0.3%) higher increase in

operating profit, and 6% (20%) higher stock return compared to their benchmark non-SOE

portfolios, by size and industry (size and market-to-book ratio), respectively. The results

suggest that the Split-Share Structure Reform contains a significant positive privatization

effect, supporting H1.

5.2 Sources of the privatization effect

This section analyzes the significance of the privatization effect and explores its sources. The

evidence shows that privatization-induced improvements to post-reform SOE performance

are positively correlated to the supportive activities of government agents, suggesting that

their incentive plays an important role in shaping the privatization effect. Post-reform sales

of state-owned shares are negatively correlated to the improvements to SOE performance,

indicating that control dilution works as a punitive mechanism to government agents who

fail to increase state-owned share value after the reform. Beside converging the interests of

the government and private investors, the market mechanism plays an effective information

discovery role in facilitating privatization embedded in the reform.

5.2.1 Significance of the privatization effect

We formally test H1, that is, whether SOEs experienced significantly stronger post-reform

improvement in performance than non-SOEs, using the regressions specified in Eq. (2). In

Table 6, the quantile regression results indicate that post-reform increase in operating revenue

is positively and significantly correlated to state ownership. The coefficients of state owner-

26

ship are 0.46, 0.51, and 0.83 for the 25%, 50% (median) and 75% quantiles, respectively. The

t-statistics show statistical significance at the 1%, 1%, and 5% levels, respectively. Graph A

in Figure 1 illustrates the estimates of the coefficients of state-ownership for regressions over

the quantile spectrum. The post-reform change in operating revenue is positively correlated

to state-ownership for all quantiles. Post-reform increase in operating profit is positively

correlated to state-ownership in the 25%, 50% (median) and 75% quantiles, significant at

the 5% level for the 75% quantile. Graph B in Fig. 1 shows that the positive correlation

between increase in operating profit and state ownership is more evident for firms with high

profit growth. Table 7 reports the regression results of stock performance measured by the

Fama-French three-factor model-adjusted stock returns. The coefficients of state-ownership

are positive and significant for the 25% and 50% (median) quantiles. Both the operational

performance and stock return results support H1 in that improvements to SOE performance

were stronger than those to non-SOE performance, highlighting a positive and significant

privatization effect.

5.2.2 Government agents’ incentives and supportive activities

We test H2, that is, privatization-driven improvements to SOE performance are positively

correlated to government agents’ supportive activities, using the regressions specified in Eq.

(3). Table 8 reports the 50% quantile (median) regression results. It shows that privatization-

driven changes in SOE operating revenue are positively correlated to the two proxies of

government agents’ supportive activities. The coefficients of the group listing dummy are

137.2, 168.5 and 144.3 for the unadjusted, size—industry, and size—market-to-book ratio

portfolio adjusted operating revenue regressions, significant at the 1% or 5% levels. The

asset injection dummy and fund raising rounds results are consistent.

Privatization-driven changes in SOE operating profit are also positively correlated to

government agents’ supportive activities. The asset injection dummy and fund raising rounds

are significant for the two privatization-driven change in SOE operating profit measures at the

27

1% level. The stock return results are also consistent with the prediction of the hypothesis.

But highly skewed and volatile post-reform stock returns during the global financial crisis

undermine the representativeness of the stock return results.

Overall, the empirical evidence supports H2 in that privatization-induced improvements

to post-reform SOE performance are positively correlated to the supportive activities of gov-

ernment agents and SOE financing activities. Since these activities capture the incentive of

the government agents running SOEs, the results support our argument that the privatiza-

tion effect is positively related to the better-aligned incentive of government agents in the

expectation of in-depth privatization.

5.2.3 Evidence from post-reform sales of state-owned shares

Beside rewarding the government agents boosting SOE performance, the Chinese govern-

ment could choose to divest in underperforming SOEs to discipline the government agents

who fail. Sales of the state-owned shares work as a punitive mechanism to government

agents by diluting their control power and jeopardizing their future promotion. We examine

whether post-reform sales of state-owned shares are negatively correlated to privatization-led

improvements to SOE performance as a robustness check for H2. We collect data on post-

reform sales of state-owned shares to private agents and measure actual privatization with

(1) number of state-owned shares sold, (2) percentage of state-owned shares sold to state-

owned shares owned, (3) percentage of state-owned shares sold to total shares outstanding,

and (4) percentage of state-shareholders involved in the sales.

Panel A of Table 9 shows that state shareholders in 160 out of 633 SOEs in our sample

sold 2.95% of state-owned shares to private agents as of October 2011.26 The comprehensive

firms had the highest percentage of firms sold state-owned shares, 28.7%, whereas the utility

firms had the lowest 14.5%.27 An average firm sold 5.13 million shares, comprising 0.44%

26The statistics could slightly underestimate the actual sales of state-owned shares, since only sales withshares exceeding 1% of total shares outstanding are required to be publicly disclosed. Since most of sales ofstate-owned shares are in large quantity, our results provide imperfect but reasonable estimates.

27A Chinese listed firm is classified as comprehensive if it is not in the utility, real estate, manufacturing,

28

of total shares outstanding, with 31% state shareholders involved. The pattern is similar

across all five industries. Privatization appeared to take place only on a small scale after the

reform. Therefore, the privatization effects on SOE performance were largely driven by the

expectation of privatization.

Panel B of Table 9 reports the correlations between post-reform sales of state-owned

shares and privatization-driven improvements to SOE operating performance. Number of

sales of state-owned shares, number of state-owned shares sold, percentage of state-owned

shares sold to total shares outstanding, and percentage of state shareholders involved are

all negatively and significantly correlated to changes in SOE performance. The results show

that the government indeed divested in underperforming SOEs after the reform, supporting

our prior that improvements to post-reform SOE performance were driven by the incentive

of government agents in increasing state-owned share value.

5.2.4 Information discovery role of the market mechanism

We look into whether the incentives of government agents and expectation of privatiza-

tion were effectively communicated through the market mechanism adopted in the reform.

Table 10 reports the 50% quantile (median) regression results in testing H3. It shows that

privatization-led improvements to SOE operating revenue and operating profit are negatively

and significantly correlated to consideration. Privatization-driven improvements to SOE op-

erating revenue and operating profit are positively correlated to private investors’ reform

plan approval rate, supporting H3. Figures 2 confirms the robustness of these findings.

The evidence implies that through the market mechanism, government agents communi-

cate their privatization-led incentives of improving SOE performance with private investors

in exchange for their support to the reform. Private investors trade off receiving higher con-

sideration at the reform stage versus benefiting from greater improvements to post-reform

SOE performance. Moreover, they tend to approve reform plans with higher approval rates

or commercial industries.

29

when stronger SOE performance is expected. Beside balancing the interests of the govern-

ment and private investors, the market mechanism plays an information discovery role in

facilitating privatization in the reform. It is an important element for the success of the

reform and privatization.

6 Conclusions

The Split-Share Structure Reform was a landmark event in China’s financial liberalization.

It converted non-tradable state-owned shares into tradable shares, enabling in-depth priva-

tization of listed SOEs. The evidence shows that SOEs experienced remarkable increases in

output and employment without sacrificing operating efficiency. Expectation of privatiza-

tion stimulated the incentive of government agents running SOEs to take quick measures to

improve SOE performance. There is no evidence that the reform improved SOE corporate

governance without fundamentally changing their ownership structure. In contrast to the

evidence found in other transitional economies that new management help improve post-

privatization firm performance, our results show that stimulating incumbent management’s

incentive with expectation of privatization also has positive effects. The market mechanism,

which helps converge the government and private investor interests, is more effective than

crude top-down orders in making privatization happen when China enters into an in-depth

reform era. Moreover, it played an important information discovery role in facilitating pri-

vatization. The positive elements of the Split-Share Structure Reform provide useful policy

implications for China’s continued economic reforms.

30

References

Allen, F., Qian, J., Qian, M., 2005. Law, finance, and economic growth in China. Journal ofFinancial Economics 77, 57–116.

Bai, C., Li, D., Wang, Y., 1997. Enterprise productivity and efficiency: When is up reallydown? Journal of Comparative Economics 24, 265–280.

Biais, B., Perotti, E., 2002. Machiavellian privatization. American Economic Review 92,240–258.

Black, B., Kraakman, R., Tarassova, A., 2000. Russian privatization and corporate gover-nance: What went wrong. Stanford Law Review 52, 1731–1803.

Bortolotti, B., Beltratti, A., 2006. The nontradable share reform in the Chinese stock market.Unpublished working paper, Bocconi University.

Boubakri, N., Cosset, J.-C., 1998. The financial and operating performance of newly priva-tized firms: Evidence from developing countries. Journal of Finance 53, 1081–1110.

Boubakri, N., Cosset, J.-C., 2002. Does privatization meet the expectations? Evidence fromAfrican countries. Journal of African Economics 11, 111–140.

Boubakri, N., Cosset, J.-C., Guedhami, O., 2005. Postprivatization corporate governance:The role of ownership structure and investor protection. Journal of Financial Economics76, 369–399.

Boycko, M., Shleifer, A., Vishny, R., 1996. A theory of privatisation. Economic Journal 106,309–319.

Brada, J., 1996. A privatization is transition—or is it? Journal of Economic Perspectives10, 67–86.

Chen, H., Wang, H., Zhou, H., 2013. Stock return volatility and capital structure decisions.Unpublised working paper, Massachusetts Institute of Technology.

Cheung, Y.-L., Rau, R., Stouraitis, A., 2006. Tuneling, propping,and expropriation: Evi-dence from connected party transactions in Hong Kong. Journal of Financial Economics82, 343–386.

Deng, J., Gan, J., He, J., 2008. The dark side of concentrated ownership in privatization:Evidence from China. Unpublished working paper, Hong Kong University of Science andTechnology.

Dewenter, K., Malatesta, P., 2001. State-owned and privately owned firms: An empiricalanalysis of profitability, leverage, and labor intensity. American Economic Review 91,320–334.

31

Ding, Y., Su, X., 2008. Implementation of IFRS in a regulated market. Journal of Accountingand Public Policy 27, 474–479.

Fama, E., French, K., 1993. Common risk factors in the returns on bonds and stocks. Journalof Financial Economics 33, 3–53.

Firth, M., Lin, C., Zou, H., 2010. Fridend or foe? The role of state and mutual fund ownershipin the Split Share Structure Reform in China. Journal of Financial and QuantitativeAnalysis 45, 685–706.

Gompers, P., Ishii, J., Metrick, A., 2008. Extreme governance: An analysis of dual-classcompanies in the United States. Review of Financial Studies 23, 1051–1088.

Grossman, S., Hart, O., 1988. One share one vote and the market for corporate control.Journal of Financial Economics 20, 323–336.

Harper, J., 2002. The performance of privatized firms in the Czech Republic. Journal ofBanking and Finance 26, 621–649.

Hou, X., Li, Q., Luo, W., 2008. Shareholder borrowing and the profitability of Chinese listedfirms. Accounting Research June, 77–84.

Huang, W., Zhu, T., 2011. Foreign institutional investors and corporate governance in emerg-ing markets: Evidence from Chinese share structure reform. Unpublished working paper,University of Hawaii at Manoa.

Huwang, C., Zhang, S., Zhu, Y., 2006. Float liquidity, speculation, and stock prices: Ev-idence from the share structure reform in China. Unpublished working paper, NanyangTechnological University.

Inoue, T., 2005. Reform of China’s split-share structure takes shape. Nomura Capital MarketReview 8, 42–59.

Jensen, M., Meckling, W., 1976. Theory of the firm: Managerial behavior, agency cost andownership cost. Journal of Financial Economics 3, 305–360.

Koenker, R., Bassett, G., 1978. Regression quantiles. Econometrica 46, 33–50.

LaPorta, R., de Silanes, F. L., Shleifer, A., 1999. Corporate ownership around the world.Journal of Finance 54, 471–517.

Li, K., Wang, T., Cheung, Y.-L., Jiang, P., 2011. Privatization and risk sharing: Evidencefrom the Split Share Structure Reform in China. Review of Financial Studies 24, 2499–2525.

Liao, L., Liu, B., Wang, H., 2011. Information discovery in share lockups: Evidence fromthe Split-Share Structure Reform in China. Financial Management Winter, 1001–1027.

32

Liao, L., Liu, B., Zhang, W., 2010. Ten-year perspective of Chinese capital markets. Unpub-lished working paper, Tsinghua University.

Lin, C., 2000. Corporate governance of state-owned enterprises in China. Unpublished work-ing paper, Asian Development Bank.

Lin, H., 2008. The costs of large shareholders: Evidence from China. Unpublished workingpaper, Columbia University.

Lin, Y., Cai, F., Li, Z., 1998. Competition, policy burdens, and state-owned enterprisereform. American Economic Review 88, 422–427.

Liu, C., Uchinda, K., Yang, Y., 2011. Controlling shareholder, Split-Share Structure Reformand cash dividends payments in China. Unpublished working paper, Kyushu University.

Lu, F., Balatbat, M., Czernkowski, R., 2008. The role of consideration in China’s Split ShareStructure Reform. Unpublished working paper, University of New South Wales.

Ma, S., Huang, Z., Xue, Y., 2005. Split-share structure, loans to large shareholders and cashdividend policy of listed firms. Accounting Research June, 44–55.

Martin, S., Parker, D., 1995. Privatization and economic performance throughout the UKbusiness cycle. Managerial and Decision Economics 16, 225–237.

Masulis, R., Wang, C., Xie, F., 2009. Agency problems at dual-class companies. Journal ofFinance 64, 1697–1727.

Megginson, W., Nash, R., Netter, J., Poulsen, A., 2004. The choice between private andpublic markets: Evidence from privatizations. Journal of Finance 59, 2835–2870.

Megginson, W., Neffer, J., 2001. From state to market: A survey of empirical studies onprivatization. Journal of Economic Literature 39, 321–389.

Megginson, W. L., Nash, R. C., Randenborgh, M. v., 1994. The financial and operatingperformance of newly privatized firms: An international empirical analysis. Journal ofFinance 49, 403–452.

Peng, S., Bewley, K., 2010. Adaptability to fair value accounting in an emerging economy: Acase study of China’s IFRS convergence. Accounting, Auditing and Accountability Journal23, 982–1011.

Perotti, E., 1995. Credible privatization. American Economic Review 85, 847–859.

Pontiff, J., Schall, L., 1998. Book-to-market ratios as predictors of market returns. Journalof Financial Economics 49, 141–160.

Sheshinski, E., Lopez-Calva, L., 2003. Privatization and its benefits: Theory and evidence.CESifo Economic Studies 49, 429–459.

33

Shleifer, A., Vishny, R., 1997. A survey of corporate governance. Journal of Finance 42,737–783.

Song, Z., Storesletten, K., Zilibotti, F., 2011. Growing like China. American Economic Re-view 101, 196–233.

Stiglitz, J., 2002. Globalization and its discontents, Norton and Company, New York.

Sun, Q., Tong, W., 2003. China share issue privatization: The extent of its success. Journalof Financial Economics 70, 183–222.

Xiong, W., Yu, J., 2011. The Chinese warrants bubble. American Economic Review 101,2723–2753.

Xu, D., Cai, Y., Xu, M., 2005. Management shareholding and corporate performance: Evi-dence from MBOs of Chinese listed firms. Management Science (in Chinese) April, 40–47.

34

  35

Table 1. Summary Statistics

This table reports the summary statistics of sample firms. Medians of variables are reported. A firm is labeled as SOE (non-SOE) if its ultimate controlling party is (not) the state. Groups P, SL, SM, SHrepresent the sub-samples of firms with zero, low, medium and high levels of state ownership measured by the ratio of state-owned shares to total shares outstanding, respectively. Net margin rate is calculated as net income divided by revenue. Accounts receivable turnover, asset turnover, and inventory turnover are calculated using operating revenue divided by average accounts receivable, total assets, and inventory, respectively. Capital expenditure is calculated as change in gross properties, plant and equipment plus change in intangible assets. On average, US$1 exchanged into RMB7.64 during the sample period.

Full Sample

By Ultimate Control By State Ownership Variable Non-SOEs SOEs P SL SM SH

Sample Number of firms 1032 399 633 146 295 296 295 Non-tradable share-to-total share ratio 61.85% 60.15% 62.51% 60.00% 58.61% 55.99% 67.00%

Size

Total assets (RMB in millions) 1,727.19 1,264.25 2,069.67 1,350.81 1,398.51 2,016.95 2,033.64 Revenue (RMB in millions) 946.41 592.31 1,236.01 562.83 693.95 1,145.39 1,225.31 Number of employees 1,688.50 1,277.50 2,086.00 1,216.50 1,416.00 1,911.00 2,156.00 EBIT (RMB in millions) 63.74 46.73 83.75 49.17 47.35 68.09 99.42

Profitability Net margin rate (%) 3.66 3.66 3.50 3.74 3.24 3.22 4.72 ROA (%) 2.26 2.08 2.50 1.80 1.82 2.22 3.18 ROE (%) 4.95 4.25 5.26 4.55 3.39 4.97 6.28

Capital Structure

Debt-to-asset ratio 0.52 0.55 0.51 0.54 0.54 0.53 0.49 Current ratio 1.16 1.14 1.17 1.15 1.11 1.17 1.19

Growth Market-to-book ratio 1.72 1.76 1.70 1.68 1.79 1.64 1.79 Capital expenditure (RMB in millions) 61.26 35.07 87.56 42.19 41.88 70.25 96.88

Productive Efficiency

Accounts receivable turnover 5.11 3.53 5.97 3.29 4.06 5.46 6.69 Asset turnover 0.56 0.48 0.59 0.45 0.51 0.56 0.61 Inventory turnover 3.90 3.26 4.32 3.13 3.77 3.90 4.78

Management Incentive

Management shareholding (‰) 0.04 0.05 0.04 0.09 0.07 0.07 0.01

  36

Table 2. Post-Reform Changes in Output, Employment, and Productivity

This table reports changes in firm output, profitability, employment and productivity three years before and after the reform. Wilcoxon signed-rank test is applied to examine the significance of changes in medians of variables. Wilcoxon Z-test is applied to examine the significance of difference in changes between groups. Proportion Z-test is used to test whether increase (decrease) odd is greater than 1. A firm is labeled as SOE (non-SOE) if its ultimate controlling party is (not) the state. Groups P, SL, SM, SHrepresent the sub-samples of firms with zero, low, medium and high levels of state ownership measured by ratio of state-owned shares to total shares outstanding, respectively. A variable with ※ is associated with percentage change calculated as the difference between its post-reform and pre-reform values normalized by pre-reform value. Otherwise, the change is calculated as the level difference between post-reform and pre-reform values. Variables in Italic are incomparable before and after the revision of the Chinese Generally Accepted Accounting Principles effective in 2007, and reported only for reference. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Full Sample By Ultimate Control By State Ownership

Increase

Odds Median Change Non-SOEs SOEs

SOEs –non-SOEs P SL SM SH

SH – P

Panel A. Output, Employment,and Capital Expenditure

Operating revenue※ 779/253***

(16.4) 0.73***

(187950) 0.57*** (24001)

0.84*** (79305)

0.27*** (3.6)

0.47*** (2832)

0.60*** (14195)

0.74*** (16499)

0.92*** (18301)

0.44*** (3.8)

Operating profit※ 681/334*** (10.9)

0.45*** (133594)

0.44*** (18187)

0.50*** (53214)

0.05 (1.4)

0.45*** (2521)

0.34*** (10260)

0.60*** (11584)

0.51*** (10449)

0.05 (1.2)

Total assets※ 850/182*** (20.8)

0.81*** (219524)

0.62*** (29966)

0.98*** (89348)

0.35*** (4.2)

0.64*** (3886)

0.61*** (17049)

0.93*** (19071)

1.13*** (19999)

0.50*** (3.5)

No. of employees※ 606/424*** (5.7)

0.13*** (93920)

0.02*** (7367)

0.19*** (47277)

0.17*** (4.3)

0.00* (968)

0.09*** (6842)

0.17*** (9500)

0.16*** (9413)

0.16*** (2.6)

Capital expenditure (%)

317/556*** (8.1)

-2.76*** (57125)

-3.87*** (9004)

-2.04*** (20297)

1.83* (1.9)

-6.28*** (1222)

-2.76*** (4955)

-3.58*** (6684)

-1.12* (2031)

5.16*** (3.1)

Panel B.Productivity

Operating revenue per employee※

703/327*** (11.7)

0.41*** (147258)

0.41*** (21209)

0.41*** (58831)

0.01 (0.0)

0.38*** (2857)

0.33*** (10390)

0.39*** (12250)

0.53*** (14621)

0.14 (1.07)

Operating profit per employee ※

579/434*** (4.6)

0.17*** (75740)

0.19*** (13209)

0.15*** (25691)

-0.04 (1.5)

0.20*** (2074)

0.14*** (5463)

0.16*** (5750)

0.18*** (6455)

-0.02 (1.0)

ROE(%) 538/465** (2.3)

0.67*** (25465)

0.72** (4539)

0.66** (8493)

-0.06 (0.7)

0.48 (354)

1.66*** (4035)

0.67 (1984)

0.10 (590)

-0.37 (0.5)

ROA (%) 507/523

(0.5) -0.06

(1655) 0.27*

(4330) -0.17

(6190) -0.44** (2.3)

0.26 (435)

0.32** (3178)

-0.07 (997)

-0.59* (2632)

-0.85* (1.7)

  37

Table 3. Post-reform Changes in Operating Efficiency and Insolvency Risk

This table reports changes in firm operating efficiency and insolvency risk three years before and after the reform. Wilcoxon signed-rank test is applied to examine the significance of changes in medians of variables. Wilcoxon Z-test is applied to examine the significance of difference in changes between groups. Proportion Z-test is used to test whether increase (decrease) odd is greater than 1. A firm is labeled as SOE (non-SOE) if its ultimate controlling party is (not) the state. Groups P, SL, SM, SH represent the sub-samples of firms with zero, low, medium and high levels of state ownership measured by ratio of state-owned shares to total shares outstanding, respectively. The change is calculated as the level difference between post-reform and pre-reform values. Variables in Italic are incomparable before and after the revision of the Chinese Generally Accepted Accounting Principles effective in 2007, and reported only for reference. Superscripts *, **, and ***denote the significance levels of 10%, 5%, and 1%, respectively.

Full Sample By Ultimate Control By State Ownership

Increase

Odds Median Change Non-SOEs SOEs

SOEs – non-SOEs P SL SM SH SH – P

Panel A. Operating Efficiency Accounts receivable turnover

806/209*** （18.7）

4.51*** (180474)

4.24*** (29016)

4.64*** (64776)

0.39 (0.7)

3.78***

(3552) 5.26*** (16888)

4.46*** (14769)

4.60***

(12981)0.82 (0.0)

Expense-to-sales ratio 714/280*** (13.8)

0.03*** (124758)

0.03***

(15848) 0.03***

(52335) 0.00 (0.4)

0.03***

(2087) 0.03*** (9548)

0.03*** (12041)

0.02***

(10229)0.00 (0.7)

Asset turnover 624/404*** (6.9)

0.08*** (71064)

0.07*** (8977)

0.09*** (29150)

0.01 (1.0)

0.04(811)

0.08*** (5546)

0.10*** (6736)

0.08*** (6741)

0.05 (1.5)

Panel B. Insolvency Risk Cash-to-total liabilities ratio

547/485** (1.9)

0.01 (1724)

0.01 (1724)

0.02 (1166)

0.01 (0.6)

-0.01 (252)

0.06*** (4146)

0.03 (973)

-0.04***

(3926) -0.03 (1.2)

Deb-to-asset ratio 696/336*** (11.2)

0.07*** (117747)

0.05*** (10654)

0.08*** (55697)

0.03*** (3.6)

0.06*** (1459)

0.03*** (5605)

0.07*** (11500)

0.11*** (13632)

0.05*** (2.6)

Current ratio 388/644*** (8.0)

-0.15*** (73164)

-0.07*** (5395)

-0.18*** (36668)

-0.11*** (2.8)

-0.10* (1327)

-0.08*** (4042)

-0.13***

(5993) -0.25***

(8930) -0.14* (1.7)

  38

Table 4. Post-reform Change in Corporate Governance

This table reports changes in corporate governance measures before and after the reform. Panel A reports the percentage of firms engaged in agency activities. Panel B reports the median ratios of the amount of money involved in agency activities to operating revenue for firms engaged in these activities. Wilcoxon Z-test is applied to examine the significance of the difference between groups in Panel B. Groups P, SL, SM, SH represent the sub-samples of firms with zero, low, medium and high levels of state ownership measured by ratio of state-owned shares to total shares outstanding, respectively. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Pre-Reform Post-Reform Full

Sample P SL SM SH SH– P

Full Sample P SL SM SH SH– P

Panel A. Percentage of Firms Engaged in the Agency Activities Related-party transaction 43.4% 39.7% 36.9% 45.9% 49.2% 9.4%

35.7% 29.9% 32.5% 39.0% 38.3% 8.4%

Related-party trans. with controlling shareholders

29.7% 29.5% 22.7% 31.8% 34.9% 5.5%

24.0% 19.4% 20.0% 25.4% 28.8% 9.4%

Lending to controlling shareholders 42.3% 37.7% 40.0% 43.2% 46.1% 8.4%

16.6% 9.7% 11.5% 19.3% 22.4% 10.8%

Panel B. Median Ratio of the Amount of Funds Involved to Operating Revenue

Related-party transaction (%)

6.92 9.90 7.60 7.76 5.84 -4.06* (1.9)

7.16 8.34 7.17 5.47 7.74 -0.58 (0.4)

Related-party trans. with controlling shareholders (%)

5.81 5.59 6.34 5.66 5.84 0.25 (0.4)

7.36 10.70 5.40 6.48 9.15

-1.55 (0.8)

Lending to controlling shareholders (%)

1.11 1.01 1.43 0.64 1.15 0.14 (0.5)

0.11 0.49 0.07 0.14 0.10 -0.39 (0.8)

  39

Table 5. The Privatization Effect

This table reports the privatization effect measured by privatization-led post-reform changes in SOE operating revenue, operating profit, and stock return, respectively. Unadjusted change in firm operating revenue (profit) is measured as the differences of the variables three years before and after the reform, normalized by pre-reform value. Unadjusted stock return is the three-year cumulative Fama-French model-adjusted return after the reform. In measuring the privatization effect, SOE operating revenue, operating profit, and stock return are further adjusted by the median operating revenue, operating profit, and stock return of a matching non-SOE benchmark portfolio by size-industry and size–market-to-book ratio, respectively. The results of Wilcoxon signed-rank tests on the significance of changes in the medians of variables are reported in brackets. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Unadjusted

Adj. by Non-SOE Size-Industry Portfolio

Adj. by Non-SOE Size-M/B Ratio

Portfolio

Operating revenue0.84*** (79305)

0.06*** (20394)

0.06*** (19462)

Operating profit 0.50*** (53214)

0.02*** (18368)

0.003*** (15299)

Stock return 1.09*** (47764)

0.06 (2867)

0.20*** (8562)

  40

Table 6. Post-Reform Changes in Operational Performance and State Ownership

This table reportsregression results of changes in firm operating revenue and operating profit on state ownership for the full sample. We follow Koenker and Bassett (1978) to apply Quantile regressions. The results for the 25%, 50% (median), and 75% quantiles are reported. OLS regression results are reported for reference. Data is winsorized at the 1% level for OLS regressions. T-statistics are reported in brackets. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Operating Revenue Operating Profit

25%Quantile 50%Quantile 75%Quantile OLS 25%Quantile 50%Quantile 75%Quantile OLS

Intercept -94.92*** (4.9)

-73.70*** (3.1)

27.69 (0.5)

24.64 (0.5)

-45.11** (2.3)

-20.04 (0.7)

96.27 (1.3)

144.87*** (2.7)

State ownership (%) 0.46*** (2.7)

0.51*** (2.7)

0.83** (2.3)

0.52 (1.5)

0.07 (0.4)

0.34 (1.5)

1.05** (2.4)

0.51 (1.3)

Non-tradable-to-tradable ratio

-0.08* (1.8)

-0.06 (1.1)

-0.19* (1.7)

-0.11 (1.0)

-0.07 (1.6)

-0.04 (0.6)

-0.20 (1.3)

-0.10 (0.8)

Log of market cap. 34.64*** (7.5)

43.55*** (7.7)

51.19*** (4.3)

33.92*** (3.6)

27.15*** (6.1)

30.93*** (4.8)

39.20*** (3.1)

27.50*** (2.5)

Regulated industry dummy

-1.46 (0.1)

-4.27 (0.3)

7.69 (0.2)

16.41 (0.6)

-2.61 (0.1)

-16.94 (0.7)

-30.74 (0.8)

-35.93 (1.0)

H-share dummy 35.07 (0.9)

15.68 (0.8)

-80.86 (1.5)

-14.42 (0.3)

34.98 (0.9)

13.33 (0.6)

-77.80 (0.6)

-11.61 (0.2)

B-share dummy -27.59** (2.0)

-51.49*** (2.9)

-42.84 (0.9)

-32.67 (1.0)

-8.82 (0.7)

-33.98** (2.1)

-45.96 (1.3)

-74.95** (2.1)

Year 2005 dummy 97.16*** (4.8)

135.37*** (5.9)

183.85*** (4.0)

150.20 (4.4)

57.13*** (3.7)

73.48*** (2.8)

99.63 (1.3)

52.87 (1.3)

Year 2006 dummy 30.57** (2.3)

34.99** (2.2)

6.79 (0.2)

11.45 (0.4) 14.20

(1.3) 16.62 (0.9)

-23.40 (0.5)

-28.90 (0.9)

Wald test for state ownership 7.37*** 7.07*** 5.06**

F-stat=5.44***

Adjusted R2=5.45%

0.15 2.35 5.87** F-stat=3.08***

Adjusted R2=2.68%

  41

Table 7. Post-Reform Stock Return and State Ownership

This table reports regression results of the Fama-French three-factor model adjusted stock return on state ownership for the full sample. We follow Koenker and Bassett (1978) to apply Quantile regressions. The regression results for the 25%, 50% (median), and 75% quantile are reported. OLS regression results are reported for reference. Data is winsorized at the 1% level for OLS regressions. T-statistics are reported in brackets. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

25%Quantile

50%Quantile

75%Quantile OLS

Intercept 138.69*** (4.6)

231.92*** (8.0)

321.67*** (11.8)

206.14*** (7.0)

State ownership (%) 0.70** (2.4)

0.54** (2.3)

-0.14 (0.4)

0.57** (2.1)

Non-tradable-to-tradable ratio -0.42*** (4.2)

-0.29*** (3.6)

-0.33*** (3.6)

-0.33*** (3.9)

Change in operating revenue 0.40 (0.4)

0.20 (0.2)

0.78 (0.3)

0.64 (1.3)

Change in operating profit 0.15 (0.2)

0.21 (0.3)

0.25 (0.3)

0.20** (2.0)

Log of market cap. -15.93** (2.4)

-28.40*** (3.8)

-28.31*** (3.2)

-35.55*** (4.6)

Regulated industry dummy -31.75 (1.5)

-45.65** (2.0)

-43.10 (1.6)

-48.14** (2.0)

H-share dummy -60.66 (1.4)

-69.08 (1.6)

9.01 (0.1)

-21.79 (0.5)

B-share dummy -122.59** (4.7)

-133.74*** (7.5)

-154.95*** (6.2)

-130.49*** (5.4)

Year 2005 dummy -43.65***

(2.6) -49.72***

(3.9) -2.43(0.1)

-8.74 (0.5)

Wald test for state ownership 5.87** 5.04** 0.12 F-stat=5.21***

Adjusted R2=6.81%

  42

Table 8. Post-Reform Changes in the SOE Performance and Government Agents’ Supports

This table reports the results of the 50% quantile (median) regressions of changes in post-reform SOE operating revenue, operating profit, and stock returns on government agents’ supporting activities. The unadjusted changes in SOE operating revenue and operating profit are calculated as the differences between the CPI-adjusted revenues and profits three years before and after the reform, normalized by pre-reform value. Size–industry and size–market-to-book ratio adjusted changes are calculated using the unadjusted changes in SOE variables minus median changes in the variables of the matching non-SOE benchmark portfolio by size–industry or size–market-to-book ratio. The unadjusted SOE stock returns are the Fama-French three-factor model adjusted cumulative stock returns three years after the reform. Size-industry and size-market-to-book ratioadjusted SOE stock returns are calculated using the unadjusted SOE stock returns minus the median stock returns of the matching non-SOE benchmark portfolio by size–industry or size–market-to-book ratio. The supportive activities proxies are individually included in the regressions. T-statistics are reported in brackets. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Unadjusted (%)

Size-Industry Adjusted (%)

Size-M/B Adjusted (%)

Panel A. Change in Operating Revenue

Group listing dummy 137.23**

(2.3) 168.50***

(3.7) 144.26**

(2.3) Wald test for group listing dummy 5.38** 13.32*** 5.35**

Asset injection dummy 120.09***

(3.9) 113.81***

(3.8) 112.47***

(3.4) Wald test for asset injection dummy 15.26*** 14.29*** 11.58***

Fund raising rounds 18.82***

(3.8) 21.61***

(4.0) 23.06***

(5.7) Wald test for rounds of fund raising 14.54*** 15.76*** 32.70*** Panel B. Change in Operating Profit

Group listing dummy 116.23**

(2.0) 96.06 (1.6)

109.94* (1.9)

Wald test for group listing dummy 4.16** 2.48 3.62*

Asset injection dummy 108.10***

(4.7) 104.27***

(3.3) 107.14***

(3.4) Wald test for asset injection dummy 21.8*** 10.72*** 11.47***

Fund raising rounds 26.67***

(5.4) 26.02***

(5.8) 30.29***

(7.6) Wald test for rounds of fund raising 28.63*** 33.67*** 58.32*** Panel C. Post-Reform Stock Returns

Group listing dummy 15.07 (0.6)

22.81 (0.9)

14.78 (0.6)

Wald test for group listing dummy 0.3 0.83 0.40

Asset injection dummy -0.54 (0.0)

2.45 (0.1)

11.70 (0.6)

Wald test for asset injection dummy 0.00 0.01 0.31

Fund raising rounds 9.01** (2.2)

6.11 (1.5)

-1.61 (0.3)

Wald test for rounds of fund raising 4.99** 2.25 0.08

  43

Table 9.Post-Reform Sales of State-Owned Shares and SOE Performance

This table reports the statistics of post-reform sales of state-ownedsharesas of October 2011. Panel A reports the numbers of firms, state-owned shares sold, ratio of state-owned shares sold to total shares outstanding, and percentage of state shareholders involved in the sales. We apply two-tail t-test to examine the statistical significance of these variables. The t-test results are reported in brackets expect for Column (2). Panel B reports the Pearson correlations between post-reform sales of state-owned shares and post-reform change in SOE operating revenue. Superscripts *, **, and *** denote the significance levels of 10%, 5%, and 1%, respectively.

Panel A. Post-Reform Sales of State-owned Shares

Industry Obs

.

Number (% in Brackets) of SOEs Sold State-owned

Shares

Avg. Number of Shares Sold per SOE (in

millions)

Avg. % of State-owned Shares Sold

to Owned per SOE

Avg. % of State-owned

Shares Sold to Outstanding

per SOE

Avg. % of State

Shareholders Involved in

the Sales

Full sample 633 160

(25.3%) 5.13*** (9.6)

2.95*** (5.8)

0.44*** (11.0)

31*** (12.7)

Utilities 69 10

(14.5%) 3.69*** (2.7)

2.66* (1.7)

0.25*** (2.5)

20*** (3.0)

Real estate 31 8

(25.8%) 4.72** (2.4)

1.24** (2.4)

0.59** (2.1)

26*** (3.2)

Comprehensive 87 25

(28.7%) 4.87*** (4.7)

2.32*** (3.2)

0.47*** (5.4)

38*** (4.9)

Manufacturing 394 108

(27.4%) 5.94*** (7.7)

3.01*** (5.0)

0.46*** (9.3)

33*** (10.6)

Commercial 52 9

(17.3%) 1.53*** (2.6)

4.91 (1.4)

0.34** (2.3)

19** (3.1)

Panel B. Correlations between Sales of State-owned Shares and Changes in SOE Operating Revenue

Number of Sales of

State-owned Shares

Number of State-owned

Shares Sold (in millions)

% of State-owned

Shares Sold to Owned

% of State-owned Shares Sold to

Shares Outstanding

% of State Shareholders Involved in

the Sales Unadjusted -0.08** -0.08** -0.02 -0.07* -0.09** Size-Industry adjusted -0.10*** -0.11*** 0.00 -0.09** -0.11*** Size-M/B Ratio adjusted -0.09** -0.11*** 0.00 -0.08* -0.10***

  44

Table 10. Post-Reform Changes in SOE Performance and the Market Mechanism

This table reports the results of the 50% quantile (median) regressions of changes in post-reform SOE operating revenue, operating profit, and stock returns on consideration and private investor reform plan approval rate. The unadjusted changes in SOE operating revenue and operating profit are calculated as the differences between the CPI-adjusted revenues and profits three years before and after the reform, normalized by pre-reform value. Size–industry and size–market-to-book ratio adjusted changes are calculated using the unadjusted changes in SOE variables minus median changes in the variables of the matching non-SOE benchmark portfolio by size–industry or size–market-to-book ratio. The unadjusted SOE stock returns are the Fama-French three-factor model adjusted cumulative stock returns three years after the reform. Size-industry and size-market-to-book ratioadjusted SOE stock returns are calculated using the unadjusted SOE stock returns minus the median stock returns of the matching non-SOE benchmark portfolio by size–industry or size–market-to-book ratio. T-statistics are reported in brackets. Superscripts *, **, and ***

denote the significance levels of 10%, 5%, and 1%, respectively.

Change in Operating Revenue Change in Operating Profit Post-Reform Stock Return Unadjusted

(%) Size-Ind.A

dj. (%) Size-M/B Adj. (%)

Unadjusted (%)

Size-Ind. Adj.(%)

Size-M/B Adj. (%)

Unadjusted (%)

Size-Ind. Adj.(%)

Size-M/B Adj. (%)

Consideration -1.06***

(3.1) -0.84***

(2.5) -1.16***

(3.2) -0.81* (1.9)

-0.72* (1.7)

-1.10*** (2.4)

0.63 (1.1)

0.72 (1.6)

0.64 (1.6)

Approval rate (%) 1.45* (1.8)

1.73** (2.0)

1.97** (2.3)

2.02** (2.0)

1.61* (1.7)

2.42*** (2.6)

0.04 (0.1)

-0.07 (0.1)

2.46** (2.2)

State ownership (%) 0.17 (0.6)

0.35 (1.2)

0.03 (0.1)

-0.22 (0.7)

0.08 (0.2)

-0.36 (1.2)

0.00 (0.0)

0.09 (0.2)

0.11 (0.3)

Controls Yes Yes Yes Yes Yes Yes Yes Yes Yes Wald test for consideration 9.79*** 6.10*** 10.52*** 3.57* 2.82* 5.97*** 1.19 0.05 2.44 Wald test for approval rate 3.15* 3.91** 5.14** 3.81** 2.83* 6.57*** 0.13 0.11 4.75**

  45

Figure. 1. Coefficients of State Ownership in Quantile Regressions

This figure depicts the estimates of the coefficients of state ownership in the quantile regressions of changes in firm operating revenue and operating profit on state ownership for the full sample. The quantile regression method follows that in Koenker and Bassett (1978). The solid line represents coefficient values within a 0-100% quantile range. The grey area represents 95% confidence interval.

(Graph A: Operating Revenue)

(Graph B: Operating Profit)

  46

Figure. 2. Coefficients of Considerationand Reform Plan Approval Rate in Quantile Regressions for Privatization-Led Improvements in SOE Performance

This figure depicts the estimates of the coefficients of consideration and private investor reform plan approval rate in quantile regressions. The dependent variables are changes in the size-industry adjusted SOE operating revenue and operating profit in capturing privatization-led improvements to SOE post-reform performance. The quantile regression method follows that in Koenker and Bassett (1978). The solid line represents coefficient values within a 0-100% quantile range. The grey area represents 95% confidence interval.

(Graph A: Size-Industry Adjusted Operating Revenue)

(Graph B: Size-Industry Adjusted Operating Profit)